CSD in Action: Contributions of Molecular Design in Drug Discovery at Roche
Reporting in the Journal of Medicinal Chemistry (2016, 59, 9, 4087–4102), Dr Martin Stahl and team from Roche Pharmaceutical Research and Early Development discuss use cases and the role of computational methods in molecular design in small molecule drug discovery.
Here we highlight those use cases where data from the Cambridge Structural Database (CSD) aided decisions on the next round of candidates to be synthesized, evaluated, and analysed.
Why Use In Silico Methods in Drug Design
The discovery of small molecule drug candidates is an iterative optimization process, consisting of designing, making, testing, and analysing. The critical phase is the design phase, where all information on previous rounds of discovery is integrated to decide on the next round of compounds to be synthesised and evaluated.
In this paper the team from Roche discuss use cases where computational methods, including data driven and experimental insights provided by the CSD, have been pivotal in the drug design phase.
Case Study 1 – Fatty Acid Binding Protein 4/5 – Targeting Specific Residue Differences in the Binding Site
Fatty acid binding proteins (FABPs) are intracellular lipid binding proteins involved in transport, and metabolism of fatty acids. The isoforms FABP4 and FABP5 have been identified as potential diabetes and atherosclerosis targets. In previously published work, two fatty acid binding proteins compounds had been identified with decent activity on FABP4 but with no selectivity against FABP3. In the current study the team at Roche worked on improving FABP4 and FABP5 activities of these compounds while strongly reducing FABP3 activity. To do so they consistently increased the size of the quinoline 2-substituent, of one of the FABPs compounds. This strategy led to a drastic increase in the FABP3 inhibition.
CSD conformational analysis data confirmed that the pseudoaxial conformation, adopted by one of the FABP candidates to fit into the selectivity pocket, was not unusual—opening the candidate pool significantly for further optimisations.
Case Study 2 – β-Tryptase – Scaffold Hopping
Human β-tryptase is a protease involved in allergic inflammatory responses in asthma. Previous published work by Sanofi-Aventis identified a series of β-tryptase inhibitors containing piperidine scaffold linkers. The Roche team set out to use these learnings and identify potent and selective β-tryptase ligands.
They used the ReCore software – built on a CSD-derived database of fragments for finding alternative chemical scaffolds – to identify new ways of connecting functional groups. Of the 200 molecules suggested, further analysis including clustering led to 20 candidates being pitched to Roche medicinal chemists.
One CSD scaffold (ref code: BUHQEQ) was selected for further investigation owing to its novelty and ease of synthesis. The scaffold-hopped molecule showed high β-tryptase inhibition.
Case Study 3 – BACE1 Inhibitors – Scaffold Hopping
Cyclic amidine BACE1 inhibitors containing a bridging biphenyl ring, an important therapeutic approach to the treatment of Alzheimer’s disease, had been previously reported by the Roche team.
Investigations into improving the physiochemical properties of the inhibitors by replacing the phenyl ring were conducted.
The ReCore software suggested a fragment from the CSD consisting of a trans-cyclopropyl ketone as a polar replacement for a phenyl ring, leading to a candidate with slightly more BACE1 inhibitor activity and significantly improved properties including increased solubility and reduced lipophilicity.
“The β-tryptase and BACE1 examples demonstrate that scaffold replacement using CSD-derived linker motifs can identify nonobvious replacements and thus open up new chemical space for projects.” explained Dr Stahl.
Conclusion
The case studies demonstrate the power of combining computational query techniques with experimental structural data and qualitative expert assessment in the molecular design phase of drug discovery at Roche.
“The CSD just continues to grow and is becoming an ever more valuable resource for us. Mining this data with clever computational methods will continue to create value in drug discovery.” concluded Dr Stahl.
Next Steps
Read the full paper – A Real-World Perspective on Molecular Design (B. Kuhn et al., J. Med. Chem. 2016, 59, 9, 4087–4102) at https://pubs.acs.org/doi/10.1021/acs.jmedchem.5b01875.
Find out more about the CSD and the software used in this research.
Find out more about Recore: A Fast and Versatile Method for Scaffold Hopping Based on Small Molecule Crystal Structure Conformations.
Learn more on scientific research using big-data insights by attending one of the frequent CCDC webinars – details and registration here.